A method of improving the operation of a processor executing a cryptographic process, by automatically detecting faults during both encryption and decryption operations by the cryptographic process, comprises segmenting the data to be encrypted and encrypting the data segments using a complex non-linear algorithm that can lead to faults; computing an output parity bit from a selected step of the algorithm for a selected data segment, based on the input value of that segment; comparing the actual output parity bit of the selected segment with the computed output parity bit for that segment; and determining whether a fault exists, based on whether the actual output parity bit matches the computed output parity bit for the selected segment.